The Involvement of DNA-Damage and -Repair Defects in Neurological Dysfunction

1. Endogenous Damage

Reactive Oxygen Species. Reactive oxygen species (ROS), which include species such as singlet oxygen atoms, hydroxyl radicals, superoxide radicals, and hydrogen peroxide, are generated as by-products of normal mitochondrial respiration,⁴ during conditions of chronic inflammation,^5,6 or as discussed below, upon exposure to certain environmental agents. These reactive chemicals are usually neutralized by cellular defense systems, which include ROS-scavenging enzymes (superoxide dismutase and catalase), antioxidant enzymes (glutathione reductase and glutathione peroxidase) and vitamins (E and C).⁷ However, in situations when the generation of ROS exceeds the cell's antioxidant capacity (i.e., during conditions of oxidative stress), ROS amass and attack macromolecules such as lipids, proteins, and DNA. In the case of DNA, ROS react with both purines and pyrimidines to generate modified bases such as 8-oxoguanine (8-oxo-dG), 5-hydroxyuracil, uracil glycol, 5-hydroxymethylthymidine, 5-hydroxymethyluracil, cytosine glycol, 2,6-diamino-4-oxy-5-formamidopyrimidine (FAPY) adenine, FAPY guanine, and thymine glycol, to name a few.¹ Other frequent lesions arising from ROS attack of the sugar phosphate backbone include abasic sites and DNA strand breaks. Such lesions, as will be discussed later, are also products of enzymatic processing of oxidized bases during repair.^8,9

Aldehydes, Estrogens, and Reactive Carbonyl Species. ROS-mediated lipid peroxidation, glucose metabolism, and estrogen metabolism are some of the processes that lead to the production of reactive aldehydes. The aldehydic by-products of lipid peroxidation, i.e., crotonaldehyde, acrolein, 4-hydroxynonenal (HNE), and malondialdehyde (MDA), form exocyclic ethano and propane base adducts in DNA that can block base pairing and are highly mutagenic.^10,11 Oxidation of sugars, such as glucose, and sugar metabolism itself creates reactive carbonyl species, e.g., glyoxal and methylglyoxal, which can produce carcinogenic base adducts such as glyoxalated deoxycytidine.^12,13 Estrogen metabolism generates metabolites such as catechol estrogens (CEs). If incompletely inactivated by endogenous processes, such metabolites can undergo oxidation to form quinones and semiquinones (CE-Q and CE-SQ, respectively). The redox cycling of quinone moieties can generate additional ROS.^14,15 CE-Q moieties are also capable of directly attacking DNA to create bulky depurinating adducts (e.g., estrogen-DNA adducts and estradiol-induced MDA adducts).

Alkylators. Another class of endogenous DNA-damaging agents includes alkylators, such as nitrosated bile salts, which are formed by bile acids that act upon N-nitroso compounds (found in nitrite-treated meat products or food),¹⁶ and S-adenosyl methionine (SAM). SAM is a natural reactive methyl-group donor for a variety of enzymatic reactions involving a range of acceptor molecules.^17,18 SAM can inappropriately alkylate DNA bases to generate products such as 7-methylguanine, 3-methyladenine, and N⁷-carboxymethylguanine, which can be mutagenic or interfere with DNA replication or transcription.^19,20 On the basis of in vitro studies, it has been estimated that SAM-mediated alkylation accounts for approximately 4000 7-methylguanine, 600 3-methyladenine, and 10–30 O⁶-methylguanine residues per mammalian genome per day. Abasic sites are also prominent intermediates of alkylation base damage; these result from enzymatic base removal during repair (see below) or from spontaneous depurination of the hyper-labile glycosidic bond between the modified base and the sugar moiety.^21–23

Intrinsic Instability of DNA. In addition to being under constant attack from endogenous sources, the chemical structure of DNA is susceptible to spontaneous decay. For instance, pyrimidines and purines endure spontaneous hydrolytic deamination; cytosines and 5-methylcytosines are particularly susceptible (100–500 cytosines are spontaneously deaminated/cell/day to form uracil).²² Purine deamination is less frequent; adenine is deaminated to hypoxanthine and guanine to xanthine at about 2%–3% the rate of cytosine deamination.^1,22 Such deamination events eventually lead to the creation of abasic sites after the enzymatic removal of the nonconventional base moiety via a repair response (see below). Apurinic/apyrimidinic (AP) sites are also frequent products of spontaneous hydrolysis of the N-glycosidic bond that links the base to the sugar phosphate backbone; such hydrolysis arises an estimated 10,000 times per human genome per day under normal physiological conditions.²² An overview of the main sources of endogenous DNA damage, examples of the lesions produced, and the measured frequencies of occurrence are presented in Table 1.^{1,10,14,22,24–30}

2. Exogenous Damage

Physical agents. Ionizing and non-ionizing radiation is the most prominent environmental physical agent that damages DNA. Exposure to radiation can occur from both natural (cosmic rays from the sun, radioactive elements in the environment, atmosphere, soil, plants and trees) and artificial (diagnostic X-rays, smoke detectors) sources. Most ionizing radiation (IR)-induced DNA damage results from reactions with hydroxyl radicals that are generated from radiolysis of water, although some damage arises from direct ionization of nucleic acid. IR-dependent DNA lesions include “simple” oxidative damage, such as modified bases, SSBs or abasic sites, and more “complex” clustered lesions (defined as those damages within a helical turn of one another) or DSBs.^31,32 Non-IR-induced DNA damage primarily involves ultraviolet (UV) radiation from the sun. The ring-structure content and the conjugated bonds in DNA bases make them prominent absorbers of both short- and long-wave UV radiation. The most important targets of UV light are adjacent pyrimidines that dimerize to form helix-distorting photoproducts, e.g., thymidine dimers.^33,34 Again, we note that radiation-induced DNA damage is not considered a major factor in the manifestation of neurological disease.

Chemical Agents. An array of genotoxic compounds are found within the environment in air, soil, water, and food. These include polluting heavy metals such as mercury, arsenic, and cadmium, which can produce ROS that create base modifications or SSBs;³⁵ food-based carcinogens such as aflatoxin, found in oilseeds, nuts, and cereals that react with DNA bases to form bulky adducts;³⁶ and air pollutants such as radon, air-borne fine particulate matter, and secondhand smoke, which can produce both oxidative and alkylative stress.³⁷ Exposure to these and other environmental hazards has been linked epidemiologically to increased incidence of cancer and, in some cases, neurological defects.^35,38–41 In addition to inadvertent environmental exposure, occupational exposure to DNA-damaging agents such as mutagenic chemicals and drugs can occur in certain industrial, laboratory, and clinical settings. Finally, crosslinking agents such as cisplatin and base analogs such as 5′-fluoro-deoxyuridine and 5′-fluorouracil are routinely used in chemotherapy to “selectively” induce DNA damage and apoptosis in rapidly dividing cancer cells.^42–44

1. Excision Repair and SSB Processing

Three major pathways function in the removal of modified bases, mismatches, and bulky adducts. These include base excision repair (BER), mismatch repair (MMR), and NER, respectively. Each of these processes proceeds via (1) removal of the substrate base, (2) formation of a SSB gap at the excision site, and (3) re-establishment of the native DNA content by pathway-specific polymerases and ligases. Because all three repair mechanisms proceed via a SSB intermediate, we discuss them together.

BER. BER functions during all stages of the cell cycle to repair non-helix-distorting base modifications, abasic sites, and various types of SSBs. Typically, the first step in BER is recognition and removal of an inappropriate base, such as one created by ROS attack, alkylation, spontaneous hydrolysis (e.g., deamination of cytosine to uracil), or nucleotide misincorporation (i.e., certain mismatches). Such removal is performed by a class of enzymes called DNA glycosylases (Figure 1). These proteins hydrolyze the N-glycosidic bond to release the substrate base and create an abasic-site intermediate.^21,45,46 Glycosylases, such as uracil DNA glycosylase (UNG [MIM 191525]) and 8-oxo-dG DNA glycosylase (OGG1 [MIM 601982]), generally target specific kinds of base damage, although these enzymes as a whole maintain an overlapping complement of base substrates.^46–50

Once an AP site has been generated, some DNA glycosylases (e.g., OGG1) possess an AP lyase activity that cleaves 3′ to the abasic residue to produce a SSB with a normal 5′ phosphate group and an atypical 3′ α, β unsaturated aldehyde.⁵¹ More commonly, however, the AP site is incised by the action of a hydrolytic endonuclease, specifically the major mammalian AP endonuclease APE1 (APEX [MIM 107748]) (also called Hap1 or Ref1) (Figure 1). Such enzymes incise the phosphodiester bond 5′ to an abasic lesion and create a SSB with an abnormal 5′–2′-deoxyribose phosphate (dRP) residue and a conventional 3′-OH terminus that can be extended by the major gap-filling polymerase, DNA polymerase β (Pol β) (POLB [MIM 174760]).⁵² APE1 also functions to remove the 3′ α,β-unsaturated aldehydic blocking groups left behind by bifunctional DNA glycosylases (see above) to create 3′-OH priming ends.

After strand breakage, and appropriate 3′-terminal clean up if needed, BER proceeds via either a short-patch or a long-patch repair reaction. In short-patch BER (Figure 1, left), DNA Pol β adds a single nucleotide and removes the 5′-dRP group, and DNA Ligase IIIα (LIG3 [MIM 600940]) (in complex with the X-ray cross-complementing 1 (XRCC1 [MIM 194360]) protein) seals the nick to complete the process.^51–58 In situations where the 5′ terminus is not a substrate for Pol β, such as with certain oxidized or reduced AP-site fragments,^59,60 long-patch BER takes place to promote strand displacement and synthesize 2–10 nucleotides (Figure 1, right). This repair reaction is performed by Pol β or Pol δ/ɛ (POLD1 [MIM 174761])/POLE [MIM 174762]) in concert with the 5′-flap endonuclease (FEN1 [MIM 600393]), proliferating cell nuclear antigen (PCNA [MIM 176740]) and DNA Ligase I (LIG1 [MIM 126391]).^{54,58,61–63}

Recent studies on BER have shown that certain 5′ or 3′ obstructive termini at SSBs are prepared for Pol β polymerization and/or nick ligation by proteins such as polynucleotide kinase/phosphatase (PNKP [MIM 605610]),⁶⁴ Aprataxin (APTX [MIM 606350]),^65,66 and tyrosyl-DNA phosphodiesterase (TDP1 [MIM 607198])^67,68 (see more below). Finally, a critical component of BER and its sub-pathway SSBR is the scaffold protein XRCC1, which recruits several key enzymatic factors, including Pol β, Ligase IIIα, TDP1, and Aprataxin, to the site of the damage.^68–75 Figure 1 summarizes the major steps of the BER pathway.

MMR. MMR functions to remove mismatches and small insertions or deletions that arise as replication errors or during recombination.⁷⁶ Mammalian MMR is initiated by MutSα (comprised of MSH2 [MIM 609309] and MSH6 [MIM 600678]) or by MutSβ (a MSH2-MSH3 [MIM 600887] heterodimer), which recognize single base mismatches and 1–2 nucleotide insertion-deletion (ID) mismatches (MutSα) and 2–10 nucleotide IDs (MutSβ).^76,77 MutLα (MLH1 [MIM 120436]:PMS2 [MIM 600259] heterodimer) is then recruited and apparently serves to coordinate other proteins, such as PCNA, at the damage site.^78,79 Subsequent steps of MMR are facilitated by the concerted actions of PCNA (strand discrimination), Exo1 (MIM 606063) (excision), replication protein A (RPA2 [MIM 179836]), Pol δ, and a DNA ligase.⁷⁶ Because none of the components of MMR have been conclusively linked to neurodegenerative disease to date, this repair system will not be discussed in further detail. For a comprehensive review of MMR, the reader is directed to reviews by Buermeyer et al. and Modrich.^76,80

NER. NER functions largely independently of the cell cycle to remove bulky DNA adducts, such as UV-induced cyclobutane pyrimidine dimers, DNA crosslinks, and certain oxidative base modifications, and is divided into two sub-pathways: global genome repair (GGR) and transcription-coupled repair (TCR) (Figure 2). As the names suggest, GGR deals with lesions present throughout the genome, whereas TCR handles DNA damage that blocks RNA polymerase elongation during gene transcription. As described in greater detail later, many of the key factors operating in NER have been associated with the cancer-prone disorder xeroderma pigmentosum (XP) and the segmental progeria cockayne syndrome (CS), which both display elements of neurological dysfunction in a sub-set of individuals.

GGR (Figure 2, left) is initiated by recognition of the helix-distorting adduct by an XPC (MIM 278720)-HR23B (MIM 600063) complex.⁸¹ This is followed by recruitment of XPA (MIM 611153), RPA, and the other components of the general transcription factor TFIIH to the lesion site. This multi-subunit assembly establishes the pre-incision complex.^82–85 TFIIH is comprised of two helicases, XPB (MIM 610651) and XPD (MIM 278730), Cdk-activating kinases (Cdk7 [MIM 601955] and cyclin H [MIM 601953]/Mat1), and DNA-binding proteins (p34 and p44), all of which serve to unwind the duplex at the lesion site and create a small bubble structure.⁸³ Unwinding is followed by dual incision on either side of the damage by XPG (MIM 278780) (incision at the 3′ bubble-duplex junction) and the XPF (MIM 278760)-ERCC1 (MIM 126380) complex (incision 5′) to release a short single-stranded DNA segment (27–39 nucleotides) that contains the lesion.^86–89 Repair synthesis at the resulting gap is carried out by Pol δ/ɛ in coordination with RPA/PCNA/RFC (replication factor C) and is followed by DNA Ligase I-mediated nick ligation.^58,83,90 Recent evidence also implicates XRCC1-Ligase IIIα in the sealing of nicks during NER, particularly in quiescent cells.⁹¹

With the exception of the initiating mechanism, TCR proceeds in a manner similar to GGR and involves many of the same protein components.⁹² Damage recognition and processing in TCR is initiated when an RNA polymerase stalls at a lesion site and thus initiates recruitment of CS proteins CSA (MIM 216400) and CSB (MIM 133540). The exact function of CSA is not known, but it is reported to translocate to the nuclear matrix in a CSB-dependent manner after UV irradiation or exposure to hydrogen peroxide. This translocation is independent of XPA and XPC, indicating that it is specific to TCR.^93,94 CSA is reportedly linked to a ubiquitin ligase complex and might play a role in ubiquitylation of the RNA polymerase during TCR by tagging the polymerase for degradation to allow for unimpeded repair. Recent evidence also suggests that CSB might be a substrate for the ubiquitin ligase complex and that it is degraded in a CSA-dependent manner during a late stage of the repair response, presumably to permit resumption of transcription.^95,96 CSB is a member of the SWI2/SNF2 family and has DNA-dependent ATPase and ATP-dependent chromatin remodeling activities. It is part of the RNA polymerase II (RNAPII) transcription machinery and recruits various NER components to the stalled RNAPII during TCR, although its precise biochemical function in the process remains unclear.^97–99 CSA and CSB have been speculated to play a role in the removal of oxidative lesions during TCR as well; however, this is currently controversial. More recent evidence implicates these proteins in global repair of oxidative DNA damage.^100–102 The detailed biochemistry of the early steps of TCR are presently being refined, but for a more complete perspective refer to De Laat et al., Mellon et al., and Van Hoffen et al.^83,92,103

2. DSBR

DNA DSBs take on two forms: two-ended breaks, generated primarily by direct attack on DNA by a physical or chemical mutagen such as IR, or one-ended breaks, created upon replication-fork collapse at sites of DNA damage (e.g., SSBs). Three major pathways function to repair two-ended DSBs: homologous recombination (HR), which entails faithful exchange between regions of homology on sister chromatids and engages the DNA-replication machinery; single-strand annealing (SSA), which involves annealing between complementary single-stranded segments of the same chromosome on either side of the DSB; and nonhomologous end-joining (NHEJ), which involves direct ligation of processed DSB ends. The one-ended, replication-derived DSBs appear to be resolved strictly by classical HR involving sister chromatids. Discussed next are the mechanistic details of these pathways. It is important to keep in mind that the process of DSBR involves dynamic interplay between phases of recognition, cell signaling (the response), and finally, break resolution.

HR and SSA. The process of HR involves exchange of genetic information between complementary regions of two homologous chromosomes (typically sister chromatids) that are aligned for crossover (Figure 3A). It is largely an error-free form of repair and hence integral to the preservation of genetic integrity.¹⁰⁴ Because accessibility of the sister chromatid is crucial for HR (strand exchange during HR is 100-fold more frequent between sister chromatids than between disparate homologous chromosome regions), this process takes place after DNA has been replicated, i.e., in the S or G2 phase of the cell cycle.

Conventional two-ended HR is initiated by recognition of the DSB by the MRN complex comprised of the 3′–5′ exonuclease MRE11A (MIM 600814), the Rad50 (MIM 604040) ATPase, and the regulatory protein defective in Nijmegen breakage syndrome (NBS1 [MIM 602667]).^105,106 This complex acts as a break sensor and recruits the protein kinase ataxia telangiectasia mutated (ATM [MIM 607585]) to the site of the damage.^105,107,108 Once recruited and activated, ATM signals to cell-cycle-checkpoint proteins, chromatin-remodeling factors, and other DNA-repair components to halt ongoing replication and execute break resolution. We emphasize that there is distinction between components that function to elicit a DNA-damage response (e.g., ATM) and those components that function in DSB resolution.

The first step in the repair of DSBs by HR is a 5′–3′ exonuclease-directed resection to produce a 3′ single-stranded DNA overhang. The exact identity of the nuclease is unknown, although studies suggest that the MRN nuclease complex might be involved as a 5′ endonuclease.^105,109 Once generated, the single-stranded overhang is resolved either by genetic exchange with a sister chromatid or by SSA (see below), both of which are HR events (Figures 3A and 3B). Strand invasion, as the name suggests, involves invasion of the 3′ end of the single-stranded DNA overhang into the region of complementarity in the intact sister chromatid (Figure 3A). This is directed by Rad51 (MIM 179617)—the human homolog of the E. coli RecA protein—which forms a nucleoprotein filament that directs homology search, strand pairing, and ultimately, invasion of the homologous chromosome.^110–112 Several Rad family members, such as Rad51B (MIM 602940), Rad51C (MIM 602774), Rad51D (MIM 602954), XRCC2 (MIM 600375), XRCC3 (MIM 600675), Rad54 (MIM 603615), and Rad52 (MIM 60392), along with RPA, are thought to play roles in assisting Rad51 in the process.¹¹³ During strand invasion (see details in Figure 3A), Rad51 promotes the formation of the classic D-loop structure (step iii) and creates a four-stranded Holliday junction intermediate,¹¹⁴ i.e., the site where the strands crossover (step iv). Subsequently, the invading strand is extended by a DNA polymerase, most likely polymerase η (POLH [MIM 603968]).¹¹⁵ Resolution of the Holliday junction is reported to occur via a Rad51C- and XRCC3 directed mechanism, leading to the formation of a crossover or non-crossover recombination product (Figure 3A, step vi).^{104,116–118} A number of nucleases and helicases, such as the RecQ family members WRN (MIM 604611) and BLM (MIM 604610), which are mutated in the premature aging syndromes Werner and Bloom, respectively, are also reportedly involved in resolving Holliday junctions during the final steps.^119,120

Another form of two-ended DSB recombination is SSA. This pathway resolves DSBs positioned between repeat sequences as shown in Figure 3B and does not require a sister chromatid. SSA is directed by Rad52 and RPA binding to the 3′ end of the nuclease-derived single-stranded 3′ overhang. If present, exposed repeat sequences upstream and downstream of the DSB are aligned, potentially creating 3′ flap or short-gap intermediates. The nonhomologous displaced 3′ tails are most likely removed by the action of ERCC1/XPF, which is homologous to the yeast RAD1 (MIM 603153)/RAD10 complex.^121,122 After appropriate nuclease and/or polymerase processing to create ligatable nicks, the break is sealed by an as-yet-unidentified DNA ligase. SSA leads to the deletion of the genetic information between the repeats, making it error-prone.

HR at a one-ended DSB functions to resolve breaks that are formed at collapsed replication forks (Figure 3C). When a progressing fork encounters an obstacle, such as a bulky adduct or a persisting SSB, on the template strand of DNA, the fork may stall and form an intermediary structure. One such structure, the so called “chicken foot,” which arises from reversion of the stalled fork (Figure 3C, step i), has been reported in bacteria, although not in mammalian cells.^123,124 Nevertheless, the intermediate structure can be processed by endonucleases and/or helicases to create a collapsed replication fork with a one-ended DSB¹⁰⁴ (Figure 3C, step ii). The end of this DSB can then undergo exonuclease resection to generate a 3′ single-stranded overhang, which is used by Rad51 and the related HR proteins to execute strand invasion (Figure 3C, steps iii and iv) and the resolution of Holliday junction(s) (Figure 3C, step v) as described above. For a more detailed understanding of the various DSBR processes, refer to the reviews by Helleday¹⁰⁴ and Szostak et al.¹²⁵

NHEJ. In mammalian cells, two-ended DSBs can be resolved not only by HR or SSA (above) but also by NHEJ, which involves direct end-to-end ligation, and at times, limited processing (Figure 3D). This form of “repair” can lead to the restoration of the original sequence or to the addition or removal of anywhere from a few nucleotides to several kilobases of DNA, making this an error-prone response. NHEJ is the major DSBR pathway in the G1 phase of the cell cycle because a sister chromatid is not needed or available for resolution.^104,126 NHEJ is initiated by the Ku70 (G22P1 [MIM 152690])-Ku80 (XRCC5 [MIM 194364]) heterodimer, which recognizes DSB ends, aligns them, protects them from excessive degradation, and ultimately prepares them for ligation. The Ku heterodimer recruits the protein kinase DNA-PKcs (DNAPK1 [MIM 600899]) (and its cofactor inositol-6-phosphate) to establish a bridging and signaling complex that serves to recruit specific repair factors.^127–131 If complementary ends are not present at the break, they can be subject to nucleolytic degradation to create single-strand overhangs with short stretches of micro-homology. This, and other end processing events, are seemingly effected by the Artemis exonuclease in association with DNA-PK, and possibly by the MRN complex, the FEN1 nuclease, or the WRN/BLM helicases.¹³⁰ When necessary, based on its interactions with Ku and the XRCC4 (MIM 194363)-LigaseIV (LIG4 [MIM 601837]) complex, polymerization of missing nucleotides is performed by DNA Pol μ (POLM [MIM 606344]).^130–132 When necessary, 3′- or 5′-blocking ends are also removed to generate the 3′ hydroxyl and 5′ phosphate termini required for ligation, for instance, by the 3′-DNA phosphatase/5′-kinase (PNKP [MIM 605610]).^{104,127–131,133,134} The XRCC4-LigaseIV tetramer is recruited to seal the DSB.^130,134 A recently identified component of NHEJ is the Cernunnos-XLF (XRCC4-like factor) protein (NHEJ1 [MIM 611290]), which associates with the XRCC4/ligase IV complex and stimulates its ligase activity in an unknown manner.¹³⁵

1. SSBR-Associated Disorders

A few spinocerebellar recessive ataxias have recently been linked to defects in the processing of SSB ends. Such disorders include spinocerebellar ataxia with axonal neuropathy (SCAN1 [MIM 607250]) and ataxia with oculomotor aparaxia type I (AOA1 [MIM 208920]). As discussed above, SSBs are common endogenous lesions, arising as intermediates of specific metabolic processes or as products of ROS attack of DNA. Ataxia with oculomotor apraxia type II (AOA2 [MIM 606002]) and a specific case of amyotrophic lateral sclerosis (ALS4 [MIM 602433]), which both arise from a mutation in the SETX gene (MIM 608465), might also involve a SSBR defect, although this has not been explicitly shown. Spinocerebellar ataxias are characterized by early to late childhood onset of motor in-coordination in speech, gait, balance, and gaze and, in some instances, mental retardation arising out of spinocerebellar axonal and post-mitotic neuronal degeneration. In this section, we discuss the above genetic disorders and their corresponding gene and repair defects.

SCAN1. SCAN1 was first identified in nine members of a Saudi Arabian family and is characterized by cerebellar ataxia and peripheral axonal motor and sensory neuropathy that resembles Charcot-Marie-Tooth disease.¹³⁶ Symptoms and pathology, which include ataxia, dystharia, mild hypoalbuminaemia, mild hypercholesterolaemia, and sensory loss, manifest around the second decade of life.¹³⁷ An interesting feature of this autosomal-recessive disorder is that the symptoms are purely neurodegenerative, without any cancer predisposition, suggesting selective assault on nondividing cells.

SCAN1 arises by a homozygous mutation in the TDP1 gene. The specific mutation (A1478G) results in substitution of the conserved histidine residue 493 with arginine in the active site of the protein.¹³⁶ TDP1 normally removes DNA-bound topoisomerase 1 (Top1 [MIM 126420]) molecules that arise during abortive enzymatic reactions and is thought to be part of the SSBR pathway.^138–140 Top1, a protein involved in the regulation of DNA supercoiling during transcription and replication, transiently breaks one strand of DNA and in the process forms a covalent link with the 3′ terminus.^67,141 After strand cleavage to relieve DNA tension, the protein-DNA intermediate is typically released, and the strand break is resealed. In the event of an incomplete reaction, which may occur when the Top1-DNA complex is formed in the vicinity of a DNA lesion, such as a modified base, gap, or nick or when cells are exposed to the chemotherapeutic agent camptothecin, the topoisomerase is irreversibly trapped on DNA, creating an obstacle for processes such as replication and transcription.^41,139 In most situations, TDP1 will resolve the abortive phosphotyrosine-DNA linkage, releasing the Top1 protein (or a degradation product) and generating a SSB that can be processed by components of BER/SSBR.¹⁴² The A1476G mutation leads to an approximately 25-fold reduction in TDP1 enzyme activity and results in the accumulation of Top1–DNA reaction intermediates.¹⁴³

Supporting the role of SSBR in Top1-DNA removal are the following observations: (1) cells defective in XRCC1, a key scaffold protein in BER/SSBR, are sensitive to camptothecin, a Top1 inhibitor,^144,145 and (2) XRCC1's protein partner, Ligase IIIα, associates with TDP1, and together the XRCC1-Ligase IIIα complex stimulates TDP1 activity at 3′-phosphotyrosine SSBs in vitro.^73,146 Recent studies by El Khamisy et al. found that SCAN1 lymphoblastoid cells exhibit a delay in the removal of IR-induced SSBs and accumulation of IR-induced Top1-DNA abortive complexes.¹⁴⁶ Additional evidence supporting a role of TDP1 in neurodegeneration comes from the studies by Katyal et al. on TDP1^−/− knockout mice. These animals show age-dependent and progressive cerebellar atrophy, and their cerebellar neurons and primary astrocytes are defective in rapid repair of SSBs associated with Top1–DNA complexes or oxidative DNA damage.¹⁴⁷ This observation, combined with the purely neurological pathology seen with SCAN1 patients, suggests that SSBs are particularly cytotoxic to nondividing cells, perhaps even more so to neurons as a result of their higher energy needs and elevated oxidative stress.

We point out that SCAN1 is a disorder of a larger family of related disorders, collectively termed spinocerebellar ataxias (SCAs), encompassing at least 29 genetic loci.¹⁴⁸ The so-called autosomal-dominant SCAs constitute a group of progressive ataxias with some similarity in their neuropathology, including olivopontocerebellar atrophy and cell loss from the Purkinje layer. Some of the SCAs are known to be caused by expansion of CAG trinucleotide repeats in the coding region of the mutated gene; such repeats lead to abnormally long polyglutamine stretches in the protein. None are conclusively or directly linked to defects in DNA repair, although there has been some indication of a repair defect in Machado-Joseph disease (SCA3 [MIM 109150]). Specifically, the defective gene product, MJD1 (or ataxin-3) (ATXN [MIM 607047]), interacts with the human homologs of the yeast DNA-repair proteins RAD23A (MIM 600061) and Rad23B (MIM 600062), HR23A and HR23B.¹⁴⁹ The dominant SCAs will not be covered further here, and the reader is directed to reviews by Schols et al.¹⁵⁰ and Paulson¹⁵¹ for additional details.

AOA1. First discovered in 1988 as a disorder with AT-like symptomology, AOA1 is characterized by early-onset cerebellar ataxia and atrophy, marked loss of Purkinje cells and peripheral nerve fibers, degeneration of posterior columns and spinocerebellar tracts, hypoalbuminaemia, and hypercholesterolaemia.^137,149 Unlike AT, yet similar to SCAN1, AOA1 does not involve non-neurological features such as cancer susceptibility and immunodeficiency.

The defective gene locus for AOA1 (APTX) was identified in members of Portuguese and Japanese families independently and was shown to map to chromosome 9p13.¹⁵² The gene product Aprataxin is a ubiquitously expressed, novel member of the histidine triad (HIT) superfamily of hydrolases/translocases.^152,153 Two alternatively spliced variants have been observed, the longer and more abundant of which is a 342 aa protein that can be divided into three domains based on sequence homology. First, the N-terminal region shares homology with PNKP, a protein involved in SSBR and known to interact with other SSBR proteins, i.e., XRCC1, Pol β and DNA Ligase IIIα.⁶⁴ Aprataxin has itself been shown to interact with XRCC1 (SSBR) and XRCC4 (DSBR/NHEJ), as well as the strand-break sensor poly (ADP-ribose) polymerase 1 (PARP1 [MIM 173870]).^72,154 Second, the central region of Aprataxin harbors the histidine triad motif typical of the HIT family members, which display nucleotide-binding and hydrolase activities.¹⁵⁵ Indeed, recombinant Aprataxin has been shown to exhibit hydrolase activity on the adenosine-5′-monophosphoramidate and diadenosine tetraphosphate substrates of the histidine triad nucleotide-binding protein (HINT) and the fragile-HIT (FHIT) subfamily of the HIT proteins.^156,157 Aprataxin was demonstrated to hydrolyze obstructive 5′ adenylate, as well as 3′ phosphate and 3′ phosphoglycolate, groups from SSBs and thereby create ligatable DNA ends.^66,158 Supporting a role for Aprataxin in SSBR, AOA1 cells are hypersensitive to agents that generate SSB damage, such as hydrogen peroxide and methymethane sulfonate (MMS), show elevated levels of DNA SSBs, and are defective in the repair of these breaks over time.^159–161 Most of the mutations discovered in AOA1 family members are confined to the catalytic domain, rendering the protein enzymatically inactive. Finally, the carboxy terminus contains a divergent zinc-finger motif that is involved in binding undamaged and nicked adenylated duplex DNA substrates.^152,156,162

AOA2. AOA2, which shares some neuropathology with AOA1, is an early-onset (10–22 years of age) disorder characterized by spinocerebellar ataxia with cerebellar atrophy, peripheral neuropathy, oculomotor apraxia, loss of Purkinje cells, mild fibrous gliosis, and elevated serum α-fetoprotein. The disease was first identified in members of Japanese, Pakistani, and Israeli families.¹⁶³ The gene defective in AOA2 (SETX) was localized to chromosome 9q34 and encodes a novel member of the superfamily 1 helicases.^163–165 Homology searches indicate that this protein, Senataxin, is an ortholog of the yeast RNA helicase Sen1P and is similar to the human helicases RENT1 (MIM 604030) and IGMBP2 (MIM 600502), both of which are involved in aspects of RNA processing.¹⁶³ Although neither helicase activity nor a specific role in SSBR has been assigned to Senataxin, we have included AOA2 under the SSBR-related disorders on the basis of the similarities in clinical and cellular phenotypes to AOA1. In particular, AOA2 cells show a similar pattern of sensitivity to the SSB-inducing agents hydrogen peroxide and MMS (but not to IR) and a higher basal level of oxidative DNA damage (8-oxo-dG).¹⁶⁶ The presence of a helicase domain suggests a role in some aspect of nucleic acid metabolism, perhaps the processing of oxidative DNA damage. Disease-associated mutations of Senataxin have been mapped in several ethnically diverse families of AOA2 patients and include missense, nonsense, and frameshift mutations, many of which lead to a truncated product.¹⁶³

Senataxin has also been implicated in a slowly progressing form of amyotrophic lateral sclerosis (ALS) called Juvenile ALS or ALS4.¹⁶⁷ ALS (or Lou Gehrig's disease) was first described by Myrianthopoulos et al.¹⁶⁸ as Charcot-Marie-Tooth disease in a family with English ancestry. It is a heterogeneous group of progressive and lethal neurological disorders that are characterized by degeneration and atrophy of motor neurons in the cerebellar cortex, spinal cord, and brain stem; the result is paralysis and eventual death by respiratory failure. Inheritance is autosomal dominant, and symptoms manifest around the second decade of life. Chen et al. identified 3 SETX missense mutations in ALS4 individuals from three geographically diverse families. The mutations—one lies in the C-terminal helicase domain of Senataxin and the two others reside in the N-terminal putative protein-protein interaction domain—are speculated to result in a partial-loss-of-function or a toxic gain-of-function protein.^169,170

2. NER-Associated Disorders

XP, CS, and Trichothiodystrophy (TTD) (TTDN1 [MIM 234050] and TTDP [MIM 601675]) constitute a spectrum of NER-related disorders broadly characterized by a varying level of photosensitivity, neurodevelopmental abnormalities, and predisposition to cancer. An interesting feature of this set of diseases is the overlap in genes involved in the manifestation of the clinical phenotypes, whereby different mutations in the same gene give rise to CS, XP, TTD, or a combination thereof. As will be discussed below, it appears that the neurological dysfunction exhibited by a sub-set of these patients arises from defective repair of specific forms of endogenous DNA damage (possibly oxidative lesions) and/or general failure of the transcriptional machinery.

XP. XP was first described in 1874 as an inherited syndrome largely affecting the skin. It constitutes a group of autosomal-recessive disorders caused by mutations in the XP gene family that participates in NER. The hallmarks of XP include sensitivity to sunlight, severe sunburns upon short exposure to UV light, excessively dry skin, freckles, dark spots, and premature aging of the skin and eyes, along with an increased risk for skin cancer. XP patients are more than 1000-fold more likely to develop cutaneous basal- and squamous-cell carcinomas at sun-exposed areas (such as the face, neck, and head) and are at increased risk for developing internal cancers by 20 years of age.^171,172 Cells from XP patients exhibit extreme sensitivity to UV radiation and poor NER after exposure to UV radiation. Although the typical features of classical XP are not related to neurodysfunction, a small fraction of XP patients (about 20%) develop neurological abnormalities, including microcephaly, progressive mental deficiency, ataxia, choreoathetosis, and areflexia.^173–176 Neuropathology in some patients includes a primary degeneration affecting large neurons in the brain, spinal cord, and peripheral nervous system and resulting in cortical atrophy, axonopathy, and gliosis.^172,175,176

The mutant genes in XP were identified as seven independent complementation groups (XPA-XPG) by fusion studies that used cells defective in various aspects of repairing UV-induced DNA damage (i.e., NER). A variant XPV group was found to be associated with an NER-proficient form of XP, but this group maintained the characteristic skin sensitivity to UV light. The genetic defect in XPV is linked to Pol η, which is required for efficient and accurate translesion synthesis (TLS) past UV photoproducts.¹⁷⁷ Cells from these patients show enhanced arrest of DNA replication at pyrimidine dimer sites and increased mutagenesis and recombination. Of the several mutations identified in POL η, the majority result in severe truncations of the protein, whereas others in the conserved catalytic domain of the protein affect the TLS activity of the polymerase.¹⁷⁸

The complementation groups XPA, XPB, XPD, and XPG have been associated with some form of neurological pathology. Certain XPB, XPD, and XPG mutations also result in a dual XP/CS phenotype that displays differing degrees of neuronal, developmental, and skin abnormalities and will be discussed in detail in the section on CS below. XPA, in combination with XPC, facilitates lesion recognition and verification during the early steps of NER and might also play a role in conferring processivity to the XPF and XPG nucleases.^179,180 Several XPA deletions or splice-site mutations that result in a frameshift have been reported in the DNA-binding domain of the protein and are associated with a neurological form of XP. Almost all of these mutations seriously disrupt protein composition and/or structure and the ability of XPA to recognize lesions and initiate NER.¹⁸¹ Several missense mutations have also been reported in the C-terminal TFIIH-binding domain of XPA and are associated with a much milder neurological phenotype. XPF mutations in humans are generally associated with milder forms of XP, although one mutation has been linked to a progeroid syndrome, which might result from roles of NER proteins, e.g., XPF, in the repair of endogenous DNA crosslinks.¹⁸² It is interesting to note that the XPF knockout mouse exhibits a severe phenotype,¹⁸³ which might imply that “mild” XPF mutations compatible with viability will be more “common” in humans. A recent case of an inherited ERCC1 defect that is associated with severe developmental failure and cerebro-oculo-facio-skeletal syndrome (COFS4 [MIM 610758]) has been reported to mimic the XPF cellular phenotypes in showing reduced levels of ERCC1-XPF and a mild NER deficiency.⁸⁴ A more comprehensive analysis of the XP gene mutations is presented by Itin et al.¹⁸⁴ and Cleaver et al.¹⁸⁵ Finally, XPC mutations have not directly been linked to a neurological phenotype, yet one case of a splice-site mutation at exon 9 in the XPC gene has been reported in connection with autism in a Korean subject.¹⁸⁶

The defects described above are largely confined to genes that modulate mainly GGR. Evidence indicates that in nondividing cells global DNA repair, particularly GGR, is mostly confined to actively transcribed gene regions, possibly as a result of DNA accessibility and an effort to conserve energy resources (for review see Nouspikel¹⁸⁷). The consequence of a repair deficit in noncoding regions is that, in the event of an unexpected checkpoint collapse, culminating in reentry into the cell cycle, attempted replication of DNA would be crippled by the accumulated lesions. This has in fact been speculated to be one of the reasons for neuronal loss in certain neurodegenerative diseases^187,188 and might also explain the neuronal phenotypes associated with certain cases of XP. However, it remains unclear how this phenomenon would impact certain brain regions selectively.

CS. First reported in 1936 by Edward Alfred Cockayne, a British physician, CS is an early-onset, progressive neurological disorder characterized by dwarfism, microencaphaly, mental retardation, sensitivity to sunlight, retinal degeneration, partial deafness, and facio-skeletal and/or gait abnormalities, but no increased cancer incidence.¹⁸⁹ In terms of its neuropathology, the CS brain shows increased fibrosis, neuronal dystrophy, and an accumulation of senile plaques and/or neurofibrillary tangles along with progressive demyelination or dysmyelination.¹⁹⁰ CS is divided into two types: type A (CSA) and type B (CSB). CSA is caused by mutations in the ERCC8 (MIM 609412) gene, and CSB is caused by ERCC6 (MIM 609413) mutations. The genes map to chromosome loci 5q12 (CSA -ERCC8) and 10q11 (CSB -ERCC6)^93,191 and fully complement the diagnostic UV-related defect in RNA-synthesis recovery in the appropriate CS cell lines.¹⁹² Two clinical variants of type B, the classic severe infantile variant and COFS syndrome, are both reported to be linked to defects in CSB.¹⁹³ Several different mutations, including missense mutations, nonsense mutations, frameshifts, insertions, splice mutations, and polymorphs have been identified in the CSA and CSB genes of an ethnically diverse group of CS patients; the vast majority of these mutations lead to a truncated protein product, and approximately 80% affect CSB.¹⁹⁴

CSA and CSB operate in TCR of UV adducts (reviewed in Laine and Egly¹⁹⁵), which explains the defect in RNA-synthesis recovery after UV exposure of CS cells; GG-NER is unaffected in CS.^196,197 CSA is a 44 kDa protein that interacts with CSB and other components of the transcription or repair machinery; for example, such components include XAB2, the p44 subunit of TFIIH, and the hyperphosphorylated form of RNAPII. CSA is needed for recruiting the nucleosomal binding protein HMGN1 (MIM 163920), the XPA-binding protein (XAB2 [MIM 610815]), and TFIIH to the stalled RNAPII and is linked to a ubiquitin ligase complex that regulates CSB degradation and RNA polymerase ubiquitylation during TCR.^{93,95,96,198,199} CSB is a member of the SWI2/SNF2 superfamily and has DNA-dependent ATPase and ATP-dependent chromatin-remodeling activities.²⁰⁰ It interacts with RNAPII in vitro and in vivo, stimulates elongation by RNAPII in vitro, and is needed for assembly of the NER proteins and histone acetyltransferase p300 (EP300 [MIM 602700]) at a stalled RNAPII.^97,98,102 Both CSA and CSB mutant cells are also hypersensitive to oxidative stress and are defective in the removal of certain oxidative lesions, such as 8-oxo-dG and (5′S)-8,5′-cyclo 2′-deoxyadenosine, suggesting that these proteins might be involved in the processing of oxidative DNA damage.^{100–102,201} Whereas 8-oxo-dG is excised primarily by the classic BER pathway, the (5′S)-8, 5′-cyclo 2′-deoxyadenosine lesion is corrected specifically by NER. Although there is no direct evidence indicating that cyclopurines are causative in neuronal cell loss, these adducts are known to block transcription and could contribute to the manifestation of the neurological disease seen in some XP and possibly some CS patients.²⁰² In addition, there is in vitro biochemical evidence that suggests that the neurodegenerative phenotype of some XP patients might stem from a role of NER in the removal and repair of classic oxidative base lesions such as 8-oxo-dG and thymine glycol.¹²⁴

It is likely that the developmental and neurological defects observed in CS (and probably XP) stem in part from the inability to efficiently process endogenous DNA damage from genes undergoing active transcription. However, there is also evidence indicating that CSB plays a direct role in regulating transcription. De Sanctis et al. demonstrated that CSB cells showed an overall defect in the recruitment of RNA Pol II and basal transcription factors after UV exposure.²⁰³ The complexity and variation of traits observed in CS patients might thus arise out of a combination of transcription and DNA-repair defects.

It is noteworthy that traits of CS are observed in a fraction of XP patients that harbor mutations in XPB, XPD, or XPG, which participate in both TCR and GGR. Such individuals are reported as XP/CS and are characterized by clinical phenotypes of both disorders. Specifically, patients with XP/CS show the sun sensitivity and skin and eye defects of XP and CS's neurological and somatic features, such as short stature and developmental abnormalities. Neuropathology of XP/CS includes cerebellar and cerebral atrophy, dysmyelination, calcification of basal ganglia, and Purkinje neuron degeneration.¹⁷⁴

XPB (or ERCC3 [MIM 133510]) maps to chromosome 2q21 and encodes an ATP-dependent 3′-5′ helicase,²⁰⁴ and XPD (or ERCC2 [MIM 126340]) maps to chromosome 19q13.2 and encodes a 5′-3′ helicase.^205,206 Both XPB and XPD are members of the transcription complex TFIIH, a central member of the transcriptional machinery, and are involved in basal transcription as well as NER. Mutations in XPB associated with CS have been reported in four families, and each leads to a truncation of the protein's C-terminal end, which harbors the critical helicase motifs. Cells from these patients have reduced XPB protein and are defective in post-UV RNA synthesis and DNA repair as measured by unscheduled DNA synthesis.^207,208 In vitro studies on two of the mutant XPB proteins immunoprecipitated from cells of XPB-CS patients indicate that defective promoter opening and reduced transcription by TFIIH might be responsible for the observed phenotypes.²⁰⁷ Additionally, one of the mutations was shown to weaken the interaction of XPB with p52, which is speculated to be a regulator of XPB ATPase activity.²⁰⁹ A tenth subunit of TFIIH, p8-TTD-A, which is mutated in a subset of TTD cases (see more below), has also been shown to regulate XPB ATPase activity and hence DNA unwinding at the lesion.^82,210 Thus, the complexity and variation of symptoms in XPB patients can probably be attributed to the disruption of one or more of these interactions. The nature of the disruption(s) would lead to variable disassembly of the TFIIH complex and in turn differentially affect transcription and/or NER responses.

XPD interacts with p44, a TFIIH core subunit that stimulates XPD helicase activity,²¹¹ and with the Cdk2 activating kinase (CAK) complex.²¹² The latter association suggests that XPD mediates binding of CAK to the core TFIIH. XPD mutations at the C-terminal end of the protein, which is the interaction site for p44, have been reported in XP/CS patients. Studies of one such XPD mutant protein revealed a drop in intrinsic helicase and Cdk7-dependent kinase activities in affected cells, leading to the hypothesis that such mutations affect TFIIH stoichiometry and result in decreased transcription and defective NER.^207,211 An N-terminal mutation has also been reported in two new XPD-CS patients. Cells from these individuals exhibit reduced levels of TFIIH, a characteristic of XP/CS,²¹³ and random DNA breakage at sites distant from the actual damage after UV irradiation, a feature previously observed in two reported XPD/CS cases and one that is presumably responsible for the extreme UV sensitivity observed in these patients.²¹⁴

An additional component reported to support CAK-TFIIH assembly is the ERCC5 (MIM 133530) gene product, XPG.²¹⁵ XPG, which maps to chromosome 13q32-33, constitutes another XP/CS complementation group. The gene encodes an endonuclease that facilitates removal of adducts by incising 3′ to the damage during classic NER.^87,216 Several missense and nonsense XPG mutations that result in inactive or severely truncated proteins have been found in a diverse set of individuals with XP/CS.²¹⁷ The truncation mutants bring about the most severe phenotype, with mortality in infancy or by early childhood. Emmert et al. correlated severity of phenotype with the extent of XPG transcript, where a milder phenotype corresponded to a higher level of mRNA. Cells harboring a severely truncated XPG with very low levels of transcript showed exceptionally poor recovery of post-UV RNA synthesis, very low post-UV cell survival, and impaired DNA repair.²¹⁷ In addition to its direct role in NER, XPG has been suggested to coordinate recognition of stalled transcription complexes during TCR in concert with CSB and TFIIH.²¹⁸ Recent studies by Ito et al. also documented an altered interaction between XPG-TFIIH and an increased disassociation of CAK and XPD from the TFIIH core in cells from either XPG or XPG-CS patients, suggesting that XPG may play a central role in stabilizing these protein complexes.²¹⁵ Finally, there is some evidence that XPG might facilitate repair of oxidative DNA damage, possibly by fostering the DNA-binding and AP lyase activities of the glycosylase hNth1 during BER.²¹⁹ The severity of phenotype in XPG patients might therefore be a combined result of the loss of excision repair, transcription, and possibly, efficient removal of oxidative DNA lesions.

TTD. TTD completes the trilogy of neurological disorders related to defects in NER. Tricho (Greek for hair)-thio (sulfur) dystrophy (faulty nourishment) aptly describes this autosomal-recessive disorder characterized by low-sulfur content, brittle hair, and variable neurological and somatic abnormalities. TTD defects include photosensitivity, ichthyosiform erythroderma (fish-like scales on skin), and progeria-like faces (faces that look prematurely aged), microcephaly, ataxia, mild mental retardation, neurodysmyelination of cerebrum white matter, and in some cases, calcification of the basal ganglia. On the basis of its clinical phenotype, TTD can be divided into two groups, a non-photosensitive group without defects in repair and a photosensitive group with defective NER. Cells from the photosensitive group show impaired removal of UV lesions, hypermutability, decreased unscheduled DNA synthesis, and reduced survival after exposure to UV light. The severity of these phenotypes varies with the location of the mutation and the gene affected.^220,221 Within the photosensitive category, three complementation groups, XPD, XPB, and TTD-A, have been reported; 95% of the reported cases belong to the XPD group.²²²

As in XP and XP/CS, a majority of the XPD mutations that give rise to TTD are point mutations leading to a single amino acid change located in the C-terminal portion of the protein, where the substitution may affect interactions with the p44 subunit of TFIIH.^207,211 Approximately one third of the XPD-TTD mutations are located within an N-terminal hotspot and lead to an R112-to-H substitution in the DNA-RNA helicase motif of the protein. Patients that are homozygous for this mutation are severely defective in DNA repair as measured by removal of cyclo pyrimidine dimers, although the clinical phenotype is reported to be mild.^223–225 As in XP and XP/CS, XPB mutations are rare in TTD. A T116P substitution has been reported in a pair of TTD siblings and is associated with defective post-UV survival, unscheduled DNA synthesis, and cyclo pyrimidine dimer removal, yet imparts a mild clinical phenotype.²⁰⁴ The third complementation group, designated TTD-A (MIM 608780), was first identified in a patient whose cells complemented all known XP groups, yet showed very low levels of TFIIH.²²⁶ The gene product of TTD-A was recently identified and encodes p8, an 8 kDa TFIIH sub-unit that contributes to the stability of the complex in vivo.²¹⁰ As noted above, p8 also facilitates duplex unwinding at the DNA lesion and recruitment of XPA. All three TTD-A mutations reported thus far, as well as the XPD-TTD mutations, result in substantially reduced TFIIH, which could originate from disruption of the p8-XPB-XPD interactions. Of the three TTD-A alterations, one maps to a region speculated to play a role in protein-protein interactions, and the others result in either the production of a protein lacking a 15 residue conserved N-terminal region or complete loss of protein synthesis (via alteration of the initiator codon ATG to ACG). Recent structure-function studies on p8 suggest that the affected region could be central to its task because deletion mutants lacking 10 or 20 amino-terminal residues were found to be incapable of restoring the NER defects and reduced TFIIH levels associated with TTD-A cells.^210,227

3. DSB-Response-Associated Disorders

Studies have found that defects in several key participants of the DSBR response are associated with various, and often related, forms of neurodegenerative disease. As discussed below, these include inherited genetic disorders arising from mutations in proteins such as MRE11, NBS1, Cernunnos-XLF, Ligase IV, and the signaling phosphatidylinositol 3-kinase-like protein kinases (PIKKs), ATM and AT and Rad3−related (ATR [MIM 601215]). Each of these proteins functions in differing aspects of DSB resolution and/or DNA-damage-checkpoint responses. It is hypothesized that the neurological dysfunction of the associated disorders arises from (i) a defect in the processing of DSBs presumably by the NHEJ pathway and/or (ii) an inappropriate DNA-damage response, quite possibly during neural development.

The circumstances under which DSBs are formed in terminally differentiated post-mitotic, nondividing cells are not clearly understood. One possibility is that DNA metabolic intermediates, oxidative modifications, and/or SSBs are clustered at specific sites within post-mitotic cells, in a fashion similar to the multiply damaged sites created by IR.^31,228,229 This could result in the generation of endogenous DNA DSBs, where even a single unrepaired DSB has the potential to be lethal, as witnessed from studies in yeast.²³⁰ A related thought is that in highly metabolic cells such as neurons, the rates of formation of endogenous DNA damage might be higher than the normal repair capacity, and these higher rates might result in a greater likelihood of the formation of clustered lesions, perhaps more so in euchromatin domains. If DSBs are indeed formed, then defects in neuronal cells in NHEJ would clearly give rise to unwanted accumulation of these deleterious and cytotoxic damages. In neuronal cells lacking DSB-response proteins, such as AT, neuronal cells might be incapable of eliciting a proper genetic response, culminating ultimately in cell death. It would be interesting to know in the future if the disorders described below exhibit global defects in general genome DNA repair, such as BER. Ongoing work on DSB generation and damage responses in post-mitotic cells will ultimately add to our current understanding of the role of DNA strand breaks in the manifestation of neurological disease.

AT and ATLD (MIM 604391). First designated AT by Boder and Sedgwick,²³¹ the disease is characterized by progressive neurodegeneration; premature aging; predisposition to cancer, particularly lymphomas and leukemias; and immunological and reproductive abnormalities. The most obvious clinical features of AT are motor incoordination (ataxia) and the appearance of clusters of blood vessels or “spider veins” (telangiectasia) on the whites of the eyes. Some of the neurological features and neuropathology include progressive degeneration of cerebellar Purkinje and granule cells and the resulting cerebellar ataxia and dysfunction, distal spinal muscular dystrophy, occulomotor apraxia, dysarthria, dystonia, choreoathetosis, and oculocutaneous telangiectasia.

The gene defective in AT, ATM, codes for a 350 kDa member of the PIKKs family of kinases involved in signal transduction and cell-cycle control.²³² The MRN complex recruits ATM to DSBs, where it acts as a sensor to activate various downstream target proteins, such as NBS1 (see below), p53 (MIM 19110), and BRCA1 (MIM 113707), which function to mitigate checkpoint responses, repair, or apoptosis.^{107,233–235} A vast number of mutations, including frameshifts, missense mutations, nonsense mutations, and splicing mutations spanning the entire length of the ATM gene have been identified in AT patients; most of these lead to a truncated protein. Cells from AT patients show radio-resistant DNA synthesis (reduced inhibition of DNA synthesis as seen in normal cells after exposure to low doses of radiation)^236–238 and are defective in activation of the G1/S or G2/M checkpoint^239–242 and the p53 pathway after radiation.²⁴³ The cancer predisposition of AT patients, therefore, undoubtedly stems from the failure of a functional, but overwhelmed, repair system to keep up with the unchecked DNA replication; this failure then leads to genomic instability in dividing cells. The neurological features of AT presumably arise from the appearance of lethal DNA intermediates, e.g., DSBs (see above), or from an inappropriate DNA-damage response.

ATLD, as the name suggests, is characterized by features similar to those of AT, without being linked to mutations in the ATM gene. This disorder was first reported by Stewart et al., who identified mutations in MRE11 in two families initially classified as AT.²⁴⁴ Individuals from these families exhibited a clinical phenotype and cellular features typical of AT; such features included increased radiosensitivity and spontaneously occurring chromosome aberrations in peripheral blood lymphocytes. MRE11 encodes a 3′-to-5′ exonuclease, which in complex with NBS1 and Rad50 initiates many forms of recombination, including NHEJ in certain circumstances, and is involved in the maintenance of telomere length.²⁴⁵ Studies have shown that MRE11 recruits and activates ATM at DSBs and regulates DNA replication at the S phase checkpoint in response to DNA damage.^107,108,246 A recurring mutation in MRE11 in ATLD is the 1714C-T missense mutation that generates a premature stop codon. Individuals from an Italian and an English family were found to bear this mutation, and in all cases, the transcript harboring the premature stop codon was subject to “nonsense mediated decay” (a process by which mRNA transcripts with a premature stop codon are marked for degradation to prevent expression of erroneous or truncated proteins that might be lethal to the cell).^244,247 These individuals, not surprisingly, are defective in expression and function of the MRN complex and show impaired ATM activity in response to IR. Interestingly, the clinical phenotype in these individuals is mainly neurological, and there is an absence of malignancy through their fourth decade of life, suggesting that the mutation is selectively detrimental to DNA-repair functions of the nervous system.

Seckel Syndrome. Seckel syndrome (MIM 210600) is an autosomal-recessive disorder caused by mutations in the ATR gene. The disorder is characterized by growth retardation, dwarfism, microcephaly with mental retardation, and a “bird-headed” facial appearance.²⁴⁸ Although there is some heterogeneity reported with respect to the defective locus, studies in two consanguineous Pakistani families in which this locus maps to chromosome 3q22.1−q24 have identified mutations in the ATR kinase as the causative defect.^249,250 O'Driscoll et al. reported an ATR splice-site mutation, which led to reduced but residual levels of normal transcript and protein.²⁴⁸ In contrast to ATM, which functions predominantly in the IR-induced DSB response, ATR is thought to be activated by multiple forms of DNA damage, including DSBs arising at stalled replication forks, as well as, possibly, IR-induced lesions. Studies have found that ATR, like ATM, can be recruited to IR-induced DSBs in an MRN-dependent manner, suggesting that there might be crosstalk and overlap between these two kinases.^251,252 Once triggered and recruited to a DSB, ATR can phosphorylate and activate some of the same downstream ATM targets, including p53, Rad17 (MIM 603139), NBS1, and H2AX (MIM 601172), a variant form of the histone H2A. The current data suggests that the neurological phenotypes of Seckel syndrome patients, much like AT patients, in all likelihood arise from an overall defect in the DNA-damage response and the recruitment of essential repair components to the site of the lesion.

NBS. NBS (MIM 251260), first identified in 1981, is an autosomal-recessive chromosomal-instability disorder characterized by microcephaly, growth retardation, immunodeficiency, and predisposition to cancer.²⁵³ It is phenotypically indistinguishable from the Berlin breakage syndrome but involves a separate complementation group. Although cells from NBS and AT patients share many similarities, e.g., hypersensitivity to IR, radio-resistant DNA synthesis, and a deficiency in the S phase checkpoint, as well as similar chromosome breakages and exchanges, the clinical phenotypes of the two diseases are distinct. In particular, NBS patients show characteristic microcephaly rarely seen in AT and do not develop ataxia or telangiectasia. The gene defective in NBS was mapped to chromosome locus 8q21, and the gene product was identified as nibrin/p95 (NBS1), which is a critical component of DSBR.^254–256 Specifically, NBS1, as part of the MRN complex, regulates repair by activating and recruiting key enzymes and kinases, such as ATM, to the site of a DSB.¹⁰⁵ The NBS1 protein has an N-terminal forkhead-associated domain (FHA) and a breast cancer C-terminal domain (BRCT), which binds to phosphorylated H2AX at DSB ends.²⁵⁷ The C-terminal portion of NBS1 contains the MRE11-binding site, and the central region bears consensus sequences for phosphorylation by ATM and ATR.^109,258,259 Thus, through these interactions, NBS1 mediates both cell-cycle control and DSBR (HR and NHEJ) and also participates in telomere maintenance.²⁶⁰

It follows that NBS1mutations that abrogate important protein-protein interactions can lead to the same systemic breakdown of the DNA-damage signaling mechanism and cell-cycle control seen in AT. A mutation found in a vast majority of NBS patients is the 657 del5 frameshift mutation.²⁵⁶ A deletion of 5 nucleotides in exon 6, it results in two different truncation products by an alternative translation mechanism.²⁶⁰ One of these products, NBS1⁷⁰, is found to associate with the MRN complex and is thought to be partially active. Consequently, patients homozygous for this mutation display a milder phenotype than individuals harboring more severe NBS1 gene alterations. Other mutations in NBS1 have been reported over diverse geographical groups, and most lead to a truncated, nonfunctional or dominant-negative protein.

LigaseIV Syndrome and NHEJ1 Syndrome. In contrast to AT and Seckel syndrome, which seemingly arise out of defective DNA-damage signaling, LigaseIV and NHEJ1 syndromes presumably stem from an actual defect in the repair of DSBs. These syndromes therefore represent a direct link between DSBR and the manifestation of neurological disease and suggest the possibility of DSBs in post-mitotic cells, even though it is unclear how they might arise. The LigaseIV syndrome is caused by mutations in the LIG4 gene.²⁶¹ As described above, LigaseIV functions to seal DSB ends during NHEJ. Several LIG4 mutations that disrupt either the ligase domain or its interaction with XRCC4 have been identified.^261,262 The clinical features of this syndrome are very similar to those of NBS and include immunodeficiency and developmental and/or growth delay. Fibroblasts from LIG4 patients maintain the pronounced radiosensitivity and impaired DSB rejoining expected of defective repair. The NHEJ1 syndrome is caused by mutations in the recently identified NHEJ1 gene that encodes the 33 kDa XRCC4-like factor (XLF), or Cernunnos. This protein is reported to promote NHEJ via its interaction with XRCC4 and LigaseIV.^135,263 Buck et al. identified mutations in the NHEJ1 gene from five patients with severe combined immunodeficiency (SCID), microcephaly, growth retardation, and sensitivity to IR. In a separate patient, Ahnesorg et al. reported a frameshift mutation that leads to a truncated protein with severely diminished or almost no expression of XLF. Cells from this patient were found to be radiosensitive and defective in NHEJ, and both of these shortcomings were corrected by reintroduction of the wild-type XLF.¹³⁵ Thus, on the basis of the LigaseIV and NHEJ1 disorders discussed here, it seems possible that DSBs are formed and can persist in nondividing cells and thus create transcription roadblocks that lead to cell death and neurodegeneration.